Re: updated uncertainty ontology

Dear all,
          I have refined classes Agent and UncertaintyNature in our 
ontology in order to be able to express the difference between human and 
computer understanding of web resources. This gives us a vocabulary for 
UIF (uncertainty interchange format) for annotation of uncertainty 
assignements for a further processing (preferably again by machines).

I will not be able to attend today telecon

Greetings Peter

Mitch Kokar wrote:
> 
> Kathy,
> 
> Here are my replies to your changes.
> 
> On Jan 7, 2008, at 11:44 PM, Kathryn B Laskey wrote:
> 
>> Mitch,
>>
>> I made a couple of changes to the uncertainty ontology. Please look 
>> them over and let me know what you think.
>>
>> I was uncomfortable with the word "random" being used as broadly as 
>> you use it.   The standard usage of the term random connotes a 
>> phenomenon that follows a statistical law. There is much ontological 
>> debate over whether randomness in this sense really exists.  Most 
>> people would not use the label random for sentences that have a 
>> definite but unknown truth-value -- such as whether Sacco and Vanzetti 
>> were guilty. Nevertheless, we can apply probability to such sentences 
>> (see the book on the Sacco and Vanzetti case by Jay Kadane and Dave 
>> Schum).  I took the liberty of changing the term to empirical on the 
>> ontology page.  I haven't changed any of the diagrams, and if I'm 
>> overruled we can go back -- but I really think this terminology is 
>> more appropriate. Then I made randomness a subclass of empirical 
>> uncertainty.  I chose this terminology because that is the term used 
>> by Morgan and Henrion (1990), which I have added to the reference 
>> list.  It is an excellent reference on uncertainty.
> 
> I like your descriptions of Empirical. This is definitely a better 
> description than what we had before for "Randomness". However, the name 
> "empirical" seems to be strange here, but if you think this is the name 
> to use, then I have no problem. The other opposite of "empirical" is 
> "theoretical". So would you say that the other types (ambiguity, 
> vagueness and inconsistency) are theoretical and not empirical? This 
> might be the case, but it's just that I am not sure.
> 
> I would prefer a different description of Randomness. You say "sentence 
> is an instance of a class" - do we need the notion of class here? Could 
> we just say that "there is a statistical law governing whether the 
> possible worlds satisfy a sentence"? Or something of this sort?
> 
>>
>>
>> I also don't think it's right to say for the case of randomness that a 
>> sentence is satisfied in one of the worlds.  An event in probability 
>> theory is a sentence that has a definite truth-value in each world 
>> (satisfies the clarity test) and is satisfied in a subset of worlds.  
>> I changed the definition to correspond to this.
> 
> Agreed (see above).
>>
>> I have issues with your definition of vagueness and ambiguity also.  
>> For ambiguity, you say a sentence can be satisfied in many worlds.  
>> Consider a sequence of 50 coin tosses, and consider the sentence that 
>> the first toss is heads.  This sentence is not ambiguous.  Its meaning 
>> is perfectly clear.  It is satisfied in 2^49 of the 2^50 possible 
>> worlds.  I looked at many definitions of ambiguity. It means open to 
>> multiple interpretations; not clearly defined. I changed the 
>> definition of ambiguity to "the referents of terms in a sentence to 
>> the world are not clearly specified and therefore it cannot be 
>> determined whether the sentence is satisfied".
> 
> I like this. In my first attempt I wanted to capture exactly what you 
> mentioned above - open to multiple interpretations. Your description 
> captures this much better.
>>
>> I also changed vagueness to "there is not a precise correspondence 
>> between terms in the sentence and referents in the world".  The 
>> prototypical example of vagueness is the concept of "tall" -- each of 
>> the possible worlds specifies a definite height, but there is no 
>> referent in the world for the term "tall."
>>
> The example of tall is very good. This is exactly what I had mind, too. 
> My intuition here points to fuzzy logic. The only problem with the 
> description now is that vagueness looks very much like ambiguity. 
> Perhaps we should make a reference to multi-valued logic here?
> 
>> I am not thrilled with these definitions, but they are the best I 
>> could do.  I don't think the original definitions were tenable for the 
>> reasons I've given.  Does anyone care to comment or make additional 
>> changes?
>>
>> I also added anchors to the wiki page, so that links can be included 
>> to the WikiWords in the uncertainty ontology.  For example, go to the 
>> Discovery or Appointment Making use cases, which have both now been 
>> annotated.  If you click on, for example, UncertaintyNature, it will 
>> take you to the place in the uncertainty ontology where 
>> UncertaintyNature is defined.
>>
>> Kathy
>>
>> On Dec 19, 2007, at 9:19 AM, Mitch Kokar wrote:
>>
>>> Hi all,
>>>
>>> In order to annotate the "buying speakers" scenario I had to extend
>>> the Uncertainty Ontology a bit. Attached is a new version. Also
>>> attached is a graphical representation of the annotation of the
>>> scenario. I will explain the details in the telecon.
>>>
>>> ==Mitch
>>>
>>> Content-Disposition: attachment;
>>>     filename=Uncertainty-v2.owl
>>> <Uncertainty-v2 1.owl><Picture 1 5.png>
>>
> 
>

Received on Wednesday, 9 January 2008 12:00:52 UTC